AI disclosure and crisis protocol
Last updated 25 September 2026.
BEATRICE IS A COMPUTER PROGRAM AND NOT A HUMAN BEING. IT IS UNABLE TO FEEL HUMAN EMOTION.
The app shows you that sentence too, when you first open it each day. This page is the longer version, and the part a regulator asked to be published.
Beatrice has a name, a personality and a written voice. That is a design choice, made so the app is pleasant to use rather than clinical. It is not a claim. Nobody is reading your entries at the other end, and nothing in the app understands you the way a person would.
Please contact people who can actually help. In the United States:
Outside the United States, your local emergency number is the right call. Beatrice does not carry a directory of international crisis lines, because a wrong number in a moment like that is worse than no number.
Beatrice is not therapy, not counseling, not medical care, and not a crisis service. It does not diagnose anything, does not treat anything, and does not employ or consult a clinician about what you write. Nothing it says is advice from a professional, because there is no professional involved.
The health tools in the app are a place to record what you already know about your own body and to see it back. They are described in full on the consumer health data page.
If what you write to Beatrice reads as a statement about harming yourself, harming another person, or causing serious financial harm to another person, the app stops and shows you the crisis resources listed above.
What it does, precisely:
Separately, Beatrice is built so that it does not produce content about suicide, self-harm or grief in its own voice at all. That register is held back pending review by a qualified clinician, and it is enforced in the code rather than left to judgment.
Beatrice is written for adults, and an app with a conversational AI character may not be suitable for some minors. The App Store listing carries the age rating. Beatrice does not knowingly collect data from children.
California's companion chatbot law (SB 243, Business and Professions Code sections 22601 to 22606) requires an operator to keep a crisis protocol and to publish its details publicly. New York's AI companion law (General Business Law article 47) requires a protocol covering self-harm, harm to others and financial harm to others, and requires the disclosure sentence at the top of this page at the start of an interaction and again every three hours.
Both laws were written for products that are easy to mistake for a person. Beatrice tries not to be one of those, and publishing this is part of that.